The story so far: Tesla billionaire Elon Musk announced early this month that his AI startup xAI would be linked to X, and that paying users on the social media platform would be able to try out the startup’s first AI chatbot Grok.
The launch of Grok comes almost a year after OpenAI released its AI chatbot ChatGPT to the public and invited users to try it out, igniting an AI race across Big Tech companies in U.S. and China to produce similar large language models (LLMs) of their own.
While Grok is a new contender, Musk’s approach of directly mixing his AI chatbot with social media raises questions about how X users’ experience on the platform will evolve after what has already been a stormy year of shock changes since Twitter’s takeover.
Why is Musk integrating his AI startup with his social media platform?
By bringing xAI to X, paying social media users will be able to interact with the Grok chabot and help contribute to its improvement by giving it more data to work with and detecting vulnerable areas. This will also allow Musk to draw more paying subscribers to X, which has seen a dramatic drop in advertising revenue under his leadership.
(For top technology news of the day, subscribe to our tech newsletter Today’s Cache)
Musk has repeatedly highlighted X’s potential to report on world events in real time, going so far as to claim that the social media app should replace established news publications. xAi and X’s integration would enable xAI’s Grok to tap into the constantly growing reservoir of posts, media, and content available on X.
Musk also posted on November 4 that a “Grok AI assistant” would be offered as part of the X Premium+ subscription package which costs $16 per month. This is slightly lower than the $20 per month ChatGPT Plus subscription. On the other hand, users can access a basic version of ChatGPT for free.
What is Grok and how does it work?
Grok is an AI model developed by Musk’s AI startup xAI and is designed to answer questions in a way that users should hopefully interpret as witty or funny. As of early November, xAI called Grok a “very early beta product” that was the result of two months of training.
“A unique and fundamental advantage of Grok is that it has real-time knowledge of the world via the X platform. It will also answer spicy questions that are rejected by most other AI systems,” said xAI in an X post on November 5.
Musk has shared alleged screenshots of user interactions with Grok, claiming that it “loves sarcasm” and has current information. By accessing information through X, Grok sets itself apart from older AI bots such as ChatGPT, which was released with a knowledge cut-off blocking it from using the latest data in its responses.
A number of existing chatbots also refuse to answer user queries requesting assistance for committing crimes, scamming people, creating malware, generating pornography, etc. It remains to be seen how Grok and xAI will address these questions and what guardrails they will put in place to safeguard users and soothe regulators’ privacy fears.
“We believe that it is important to design AI tools that are useful to people of all backgrounds and political views. We also want empower our users with our AI tools, subject to the law,” said xAI on its site.
A prototype LLM (Grok-0) was trained with 33 billion parameters, according to xAI’s website, while the Grok-1 LLM is much more powerful. xAI said that Grok-1 had at times beaten rival models including LLaMa 2 70B, Inflection-1, and GPT-3.5 when it came to some math and code-related machine learning benchmarks.
Only a limited number of U.S.-based users will be able to access Grok at present. Once it is out of beta, X Premium+ (paying) users will be able to access the system, Musk said on November 4. Other interested applicants can join a waitlist. Grok will be built into X but will also exist separately.
How might the integration affect users?
On November 7, X CEO Linda Yaccarino shared a post which noted that “X’s new friend” Grok would help with product development, brand safety, search, and other features. AI-enabled personalisation would also be possible.
While access to Grok is still highly restricted, tracking the launch of new AI features on platforms like Google, Meta, and OpenAI can help predict the chain of future events. While many eagerly tried out OpenAI’s text-to-image generator DALL-E, Meta’s generative AI ad tools, and Google’s Bard chatbot, other users hit the three companies with copyright infringement lawsuits. Authors, journalists, and artists have claimed that the tech giants scraped their copyrighted works to train their respective AI models without seeking permission or paying the original creators.
Similarly, X users who are concerned about their work being scraped for AI training may choose to reduce or stop sharing content on the site. Others may decide to protect their work legally.
OpenAI has defended the use of copyrighted work for innovative purposes, while companies like Google and Microsoft are preparing to legally support their users who may be sued for similar reasons.
As the partnership continues, xAI and X will soon have to outline their own approach to handling copyrighted media and the ways in which they will support clients.
What is Musk’s stance on AI?
Musk’s relationship with AI is complex, and often contradicts statements he has made in the past. In March, he was one of the tech leaders who signed an open letter on the Future of Life Institute website to call for a six month pause in training new AI systems so that regulation could catch up. However, he soon announced the formation of his own AI startup xAI, claiming that it aimed to “advance our collective understanding of the universe.” Other senior xAI team members previously worked at DeepMind, OpenAI, Google Research, Microsoft Research, Tesla, and the University of Toronto.
Musk is publicly critical of OpenAI, claiming that the company’s tech is controlled by Microsoft and that the ChatGPT-maker has become a closed entity which lacks virtues and transparency. It is worth noting he was an early investor in the company who left in 2018.
xAI was announced in July this year.